Generating CCSS correction

Home Forums Help and Support Generating CCSS correction

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #38475

    Slawo
    Participant
    • Offline

    Hi,

    I am new to calibration topic and recently I have bought Calibrite ColorChecker Display Plus colorimeter. I had issues with choosing appropriate correction for my monitor (LG 27GN880). My friend borrowed me old (2006) X-rite I1 pro spectro for reference and ccss generation. Because of that I have three questions:

    1. Do RGB controls on monitor influence ccss generation measurements? If yes, should I use equal gain on all RGB channels?
    2. After creation of CCSS I am still getting around 200 K difference on white point measurements between colorimeter and spectro. Measurements are close to the one made with bundled “Panasonic VVX*” correction. When using CCMX it seems to be close to spectro readings. Is that expected?
    3. Contrast and black point readings when using spectro are unstable – between 1:1600 to 1:5000, where my monitor is max 1:1000. I have tried black level drift compensation but it doesn’t help. Anybody has an idea why this is happening?

    i1Basic Pro 2 on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #38478

    Vincent
    Participant
    • Offline

    Hi,

    I am new to calibration topic and recently I have bought Calibrite ColorChecker Display Plus colorimeter. I had issues with choosing appropriate correction for my monitor (LG 27GN880). My friend borrowed me old (2006) X-rite I1 pro spectro for reference and ccss generation. Because of that I have three questions:

    1. Do RGB controls on monitor influence ccss generation measurements? If yes, should I use equal gain on all RGB channels?

    They should not mater, it’s gains in SPD height for RGB displays

    1. After creation of CCSS I am still getting around 200 K difference on white point measurements between colorimeter and spectro.

    CCT is useless. Measure distances in dE, or d-uv or something  meaningful.

    1. Measurements are close to the one made with bundled “Panasonic VVX*” correction. When using CCMX it seems to be close to spectro readings. Is that expected?

    CCMX = match (under some tolerance) the readings from reference device (whateever it is and whatever actual accuracy those readings have).
    CCSS = use colorimeter firmware data as reference , to be corrected with some spectral power distribution (CCSS). The further colorimeter is in some wavelength interval from std observer the more important CCSS accuracy is.

    If “reference device” was an actual reference CCMX is prefered because you get a match to actual color coordinates. For example a JETI Specbos.

    If not, specially this case if it is a WLED PFS with this narrow peaks far from i1proX resolution and i1d3 is well preserved, CCSS may be able to correct the tiny deviations of colorimeter observer(firmware data) vs std observer.

    “Panasonic VVX*” is a WLED PFS phosphor 1nm CCSS sample, beyond an i1Pro capabilities, more accurate… but a custom 3nm CCSS would be better if main blue or green wavelength is “moved” from  ““Panasonic VVX*”” sample. Plot both CCSS using display cal (i) button or specplot command in argyllcms.

    1. Contrast and black point readings when using spectro are unstable – between 1:1600 to 1:5000, where my monitor is max 1:1000. I have tried black level drift compensation but it doesn’t help. Anybody has an idea why this is happening?

    Maybe dynamic contrast is on in display (it should be off) or i1Pro pushing its limits in low light.

    • This reply was modified 2 weeks, 6 days ago by Vincent.
    • This reply was modified 2 weeks, 6 days ago by Vincent.
    #38486

    Slawo
    Participant
    • Offline

    Firstly, thank you for very detailed answer.

    They should not mater, it’s gains in SPD height for RGB displays

    That is what I thought, so the important thing in this distribution how actual color is distributed over wavelengths, not actual gain itself.

    CCT is useless. Measure distances in dE, or d-uv or something  meaningful.

    Wouldn’t dE be off if correction is not accurate?

    Plot both CCSS using display cal (i) button or specplot command in argyllcms.

    I did and they do differ a little, although I am not sure if this due to some approximation/interpolation. Obviously red peaks are really cut in 3nm. I am attaching ccss file as reference. It seems close to Macbook Retina version of correction. Assuming I am using it for gaming/movies/office work, is possible error something I should consider? Which one of those corrections would you choose? Unfortunately 1nm spectro is way beyond my capabilities 😉

    Maybe dynamic contrast is on in display (it should be off) or i1Pro pushing its limits in low light.

    Dynamic contrast is off, so I guess second thing.

    Attachments:
    You must be logged in to view attached files.
    #38489

    Vincent
    Participant
    • Offline

    Firstly, thank you for very detailed answer.

    They should not mater, it’s gains in SPD height for RGB displays

    That is what I thought, so the important thing in this distribution how actual color is distributed over wavelengths, not actual gain itself.

    Yes, given a CCSS at certain xy white, all in gamut colors  can be measured… and that set includes cooler or warmer whites.

    CCT is useless. Measure distances in dE, or d-uv or something  meaningful.

    Wouldn’t dE be off if correction is not accurate?

    Against a “true” reference:

    CCMX will be off

    CCSS may be off but only if colorimeter oberser (firmware) drifts from std observer in the places where CCSS is innacurate.
    Of course if colorimeter observer is not accurate (firm observer vs actual filter response), this will add more inaccuracies on those wavelengths.
    But we are assuming that under some tolerance:
    -i1d3 does not match cie 1931 2 degree observer (OK even this causes innacurate results when uncorrected)
    -i1d3 firmware matches actual colorimeter filters response => this is the source of accuracy of a CCSS based solution.

    Plot both CCSS using display cal (i) button or specplot command in argyllcms.

    I did and they do differ a little, although I am not sure if this due to some approximation/interpolation. Obviously red peaks are really cut in 3nm. I am attaching ccss file as reference. It seems close to Macbook Retina version of correction. Assuming I am using it for gaming/movies/office work, is possible error something I should consider? Which one of those corrections would you choose? Unfortunately 1nm spectro is way beyond my capabilities 😉

    Looks closer to Panasonic. Native green is less saturated than P3, like in that bundled ccss

    Maybe dynamic contrast is on in display (it should be off) or i1Pro pushing its limits in low light.

    Dynamic contrast is off, so I guess second thing.

    #38495

    atagunov
    Participant
    • Offline

    Wouldn’t dE be off if correction is not accurate?

    Hi, definitely with our budget tools there will be inaccuracies. The way I read Vincent’s first post however is that instead of saying

    After creation of CCSS I am still getting around 200 K difference on white point measurements between colorimeter and spectro

    it might make more sense to tell us how much of a dE difference that was

    Assuming I am using it for gaming/movies/office work, is possible error something I should consider?

    So long as you’re not sending movies color graded by yourself to directors for approval a moderate error should be fine. I read somewhere that the point of calibrating those monitors that professional colorists use is rather simple: imagine an average Joe has got a horrible TV, that TV makes everything green. However so long as all the colorists around the world are using correctly calibrated displays all the movies that Joe watches will be equally green. That will be the kind of green he is used to. My point is that we have been looking at uncalibrated monitors until now and we were fine, we got used to them. There is no doubt that you can make movies look better by making your system more color accurate, but so long as you’re not doing professional photo or video work moderate inaccuracies should be fine.

    • This reply was modified 2 weeks, 5 days ago by atagunov.
Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS